The Sorting Hat Effect: Avoiding the Hazards of Endgame Scoring

I've been playing Ascension on iOS for a few months and in that time I've had a lot of fun... Until the endgame, wherein I discover all my efforts have been in vain as my opponent has doubled or tripled my score. I can deal with consistent loss in a game, but consistently losing when I think I will win is frustrating.

This speaks to the value and purpose of endgame scoring in general, I think. As all modern gamers would recognize, hidden endgame scoring keeps all players engaged until the very end. There's always this chance that you'll beat the odds, because you've had a clever engine building for the whole game. There is a lot of dramatic fun in pulling back the curtain to reveal your grand idea, even if another player ends up beating your score.

When designed well, these endgame mechanics can be learning experiences for players to try again with a slight adjustment to their strategy. However, they can also appear to be black boxes, capriciously deciding a victor after the fact.

I call this the Sorting Hat effect, after the hat who decides the dormitory for each student at Hogwart's regardless of their input. Richard Garfield uses "Randochess" as an example of this phenomenon. In Randochess, players play a full game of Chess, then roll a die. If it results in a 6, the loser of the chess match actually wins the whole game. Why does this suck? The winner of Randochess doesn't feel like they earned it and the loser feels like they were cheated.

Here are some other examples of endgame scoring:

• A lot of people dislike Carcassonne because the farm scoring is so opaque that they get blind-sided when a more experienced player wins the game thanks to some well-placed farmers. But at least in that case, the information is public, and, more importantly, can be manipulated mid-game.
• In the case of Lords of Waterdeep, each player gets an endgame bonus for having accomplished tasks in certain categories. That can have very drastic swings in endgame, but they very rarely comprise more than a third of a player's final score, so it still feels like you have control over your fate to some extent.
• In the case of Seasons, I also lose pretty handily every time, and the endgame score can be almost half of a player's final score. However, there is a lot of fun happening mid-game by building engines, timing actions, and interfering with opponent's engines.

So, this indicates a rule of thumb for myself when I design an endgame scoring mechanic.

• Visible: The endgame state should be visible to all players, even if it is a little complicated for a newcomer to decipher.
• Adjustable: The endgame state should be adjustable mid-game, so clear leaders can be recognized and targeted accordingly.
• Small: The endgame state should comprise about a third of a player's final score. More than that makes the game too swingy, less makes it feel like an afterthought.

I can deal with an endgame score that follows at least two of those three rules of thumb. (ex: If it's too big, I can deal with that because I saw it coming and could have adjusted for it.) Those are my takeaways anyhow. What are yours?